Signed preKey correspondent from SignalProtocol

[alinradut]

Hello! I'm trying to get Olm to work with XMPPFramework, but I am hitting some walls here related to compatibility between the OMEMO implementation in the framework (which seems to follow SignalProtocol's lead) and Olm.

The DoubleRatchet spec from OWS mentions a signed prekey that needs to be uploaded , however, in Olm 2.2.2 I was unable to find any reference to something similar. Does Olm support it? What would be its correspondent?

[itamarwe]

As far as I understand, Olm does not support the signed prekey. What is implemented in Olm is the handshake as described here: https://signal.org/blog/simplifying-otr-deniability/

If you want to implement Omemo using olm you can use stubs for the signed prekeys, and just ignore them when calling the Olm functions.

In terms of security, it means that you won't enjoy the security benefits of using signed prekeys. If you look at the list of attacks here, you can see what attacks the signed prekeys protect you from. Personally, I think that the signed prekeys protect from relatively marginal attacks, but that depends on your own usecase and security concerns.

[richvdh]

you can sign your prekeys if you want. See https://git.matrix.org/git/olm/about/docs/signing.rst#signing-one-time-keys

[moyerr]

@richvdh What does this look like in terms of code? OLMAccount has a method for signing a message, but it isn't clear how one would go about signing a preKey.

[richvdh]

just sign the message you are using to distribute the prekeys. You can read about how to do iit in the Matrix protocol at https://matrix.org/docs/guides/e2e_implementation.html.